shown the estimated net yearly savings
The Hydrogen-Cooled Turbine Generator
D. S. SNELL
ASSOCIATE AIEE
H YDROGEN cooling for electrical
machinery was first proposed to the
industry through an AIEE paper pre
sented in 1925. * The first hydrogen-
cooled machine to be placed in commer
cial service was a 12,500-kva synchronous
condenser, installed in 1928.2 This was
followed by other machines of this type,
and later, the hydrogen-cooled frequency-
converter was introduced. There are
now in operation in the United States 20
of these two types of hydrogen-cooled
machines, with a combined output of over
one-half million kva, and their records of
performance have been highly satisfac
tory. Prior to 1937, the application of
hydrogen cooling to turbine generators
had been confined to developmental
machines in the manufacturers' plants,
although several air-cooled generators
had been built with provision for later
adaptation to hydrogen cooling. In
October 1937 the first hydrogen-cooled
generator built for commercial service was
placed in operation at Dayton, Ohio.
This was a 3,600-rpm unit, of General
Electric manufacture, rated at 31,250
kva at 0.8 power factor. Since then, a
total of 10 hydrogen-cooled generators
have been placed in service in various
parts of the country, and 27 others are in
the course of installation or construction.
The total capacity of these 37 generators
is over 2,000,000 kva, and their sizes
range from 17,000 kva to 81,250 kva at
3,600 rpm, and from 75,000 kva to 176,-
470 kva at 1,800 rpm.
Two papers have already been pre
sented before the Institute, in which the
development and design work of one large
Paper 39-105, recommended by the AIEE commit
tee on electrical machinery, and presented at the
AIEE North Eastern District meeting, Springfield,
Mass., May 3-5, 1939, and at the combined summer
and Pacific Coast convention, San Francisco,
Calif., June 26-30, 1939. Manuscript submitted
March 3, 1939; made available for preprinting
April 1, 1939.
D. S. SNBLL is electrical engineer in the a-c turbine
generator engineering department of the General
Electric Company, Schenectady, N. Y.
The development of the hydrogen-cooled generator
described in the foregoing was carried out under the
direction of M. A. Savage, and in collaboration
with Chester W. Rice, E. H. Freiburghouse, and
Robert Hannah. The electrical system of the
hydrogen control cabinet was designed by E. J.
Flynn and A. J. Bialous. The tests on the Logan
and Cincinnati generators were obtained through
the co-operation of the Appalachian Electric Power
Company and the Cincinnati Gas and Electric
Company, which co-operation is hereby gratefully
acknowledged.
1. For all numbered references, see list at end of
paper.
JANUARY 1940, V O L . 59
with hydrogen cooling for different sizes
of generators, corresponding to these in
creases in efficiency. These savings as
sume a power cost of 0.3 cent per kilo
watt-hour and an operating time of 80
per cent and take account of the cost of
the hydrogen and carbon dioxide re
manufacturing company in connection
with the hydrogen-cooled generator have
been described.3,4 It is the purpose of
quired for operating and filling. They
the present paper to describe the work of ,Qi+Q*)X OF Ha
another large company in the commercial
development of this new-type machine,
and to present some of the results of sev
eral months' operating experience with
eight of these machines.
Advantages of Hydrogen
as a Cooling Agent
These have been stated many times be QlA(l-Z) OF AIR
fore, and are here mentioned only in the
Figure 2 . Schematic diagram of shaft-sealing
interest of completeness.
system for hydrogen-cooled generators, using
1. Windage and ventilation losses in hydro vacuum-treated oil
gen are only about one-tenth of their value
in air, due to the lower density of hydrogen.
2. The rating of a machine can be in also assume the installed cost of the hy
creased approximately 20 per cent for a given drogen-cooled generator to be the same as
amount of active material, by operation in that of the air-cooled generator, which is
hydrogen instead of in air, due to the supe usually the case. The savings thus cal
rior cooling properties of hydrogen.
culated are shown to range from about
3. The life of the winding insulation is in $4,000 a year for a 25,000-kw 3,600-rpm
creased, due to the absence of moisture,
oxygen, and dirt, and to the fact that corona unit to about $20,000 a year for a 160,-
damage is eliminated, in the hydrogen- 000-kw 1,800-rpm unit.
cooled machine. The greater output per pound of mate
4. Thefirehazard is eliminated. rial obtained with hydrogen cooling, ex
tends the limit of output of generators of
The reduction of the windage and ven the 3,600-rpm type and allows the build
tilation losses of a turbine generator with ing of generators with self-ventilation in
hydrogen cooling increases the full-load capacities which, if built for air cooling,
efficiency about 0.7 per cent for 1,800- would require external fans. The larg
rpm machines and about 0.9 per cent for est capacity at 3,600 rpm now considered
3,600-rpm machines. In figure 1 are practicable with air cooling is about 62,-
500 kva, and such a machine would have
20i-
l i t external fans; with hydrogen cooling,
E S T I M A T E D NET
SW
AV
generators as large as 81,250 kva at 3,600
I TI N
H GHYDR
PEROGEN YEAR
CO O L I NIG rpm are practicable, ventilated with in
6
^P ternal fans.
< fi Type of Shaft Seal Used
_
oi2h
Q / / The most important part in the de
/> velopment of the hydrogen-cooled genera
4
z tor was the design of a means for sealing
4 the rotating-shaft extensions against the
outward leakage of hydrogen. After in
7 vestigating many different sealing ar
rangements, of the mechanical, combina
tion mechanical-and-liquid, and liquid-
film types, the General Electric Company
% finally adopted the liquid-film type of
0 20 40 60 O IOO 120 I 4 0 160
GENERATOR RATING -THOUSAND KW AT OS P-F seal for use with its hydrogen-cooled gen
erators. This type of seal is illustrated
Figure 1 . Estimated net yearly saving by
using hydrogen in place of air as the cooling schematically in figure 2 and includes a
agent, for generators in different sizes sealing ring surrounding the shaft, con-
SnellHydrogen-Cooled Generators TRANSACTIONS 35

Figure 3. A 6,250-kva 3,600-rpm generator
built for operation in hydrogen. Coolers are
located beneath the rotor. Vacuum tank
for treating shaft-sealing oil shown in back
ground
structed integral with the bearing, to
which vacuum-treated lubricating oil is
supplied under pressure.5'6 This creates
a film of oil completely surrounding the
shaft which prevents the escape of hydro
gen from the casing at this point. This
type of shaft seal has the advantages of
simplicity and ruggedness, and through
operating with vacuum-treated oil both
permits a high degree of hydrogen purity
to be maintained in the generator casing
and prevents the contamination of the oil
in the bearing lubrication system by the
hydrogen.
Tests on a 6,250-Kva
Hydrogen-Cooled Generator
In 1926 a 6,250 kva, 3,600-rpm hydro
gen-cooled generator was built by the
General Electric Company for develop
mental purposes. This machine, figure
3, was provided with a heavy, cylindrical
casing, cast-steel end shields, and with
two gas coolers, contained in a rectangu
lar box beneath the rotor. A shaft seal
of the oil-film type was provided in each
generator bearing. This machine was
tested as a synchronous condenser in hy
drogen and in air over a period of about 12
months, during which time the operating
characteristics of the seals and auxiliary
equipment could be observed and infor
mation on the heating of the machine in
hydrogen and in air obtained. The more
important of the heating data obtained
in the tests are summarized in figures 4
to 6.
Figures 4 and 5 give the field and ar
mature heating characteristics of the
6,250-kva generator for different loss in
tensities on the coil surfaces, with air,
carbon dioxide, and hydrogen as the cool
ing media, the tests in hydrogen having
been conducted at three different hydro
gen pressures. Curves for cooling with
36 TRANSACTIONS
160r
helium have been added, based on calcu
lation. The armature heating curves in
figure 5 have had deducted the calculated
temperature rise of the gas in the air gap
due to field PR loss. This rise is of the
order of from three to five degrees centi
grade. This was done in order to use the
curves for estimating the armature tem
perature rise at 0.8 power factor loads,
the tests having been made at zero power
factor.
The temperature difference between
armature copper and between-coil tem
perature detector for different armature
loadings, in air and in hydrogen, is shown
in figure 6. At normal load, correspond
ing to a loss intensity of about 0.3 watt
per square inch, this temperature differ 0 0.5 1.0 1.5 20 2.5 3.0
W A T T S PER SQ N OP FIELD COIL
ence is about 15 degrees centigrade in air
as compared with 11.4 degrees centigrade Figure 4. Field heating characteristics of a
in hydrogen at 0.44 pound per square 6,250-kva generator with respect to tempera
ture of gas leaving coolers, from tests with
inch pressure. The lower values of the
generator operating in different cooling gases
temperature difference at higher hydro
gen pressures indicate a slight increase
in the thermal conductivity of the insula
same short-circuit ratio at all ratings (by
tion at the higher pressures. The cross
maintaining the same stator core dimen
ing of the curves at the higher armature
sions but varying the rotor diameter) the
loadings may be due to changes in the
increases in output with hydrogen cooling
insulation occurring at the higher tem
over the normal air rating, for the same
peratures.
winding temperatures, would be approxi
From the foregoing tests, and tests
made on stationary models, it was deter mately 20, 30, and 35 per cent for hydro-
mined that the thermal conductivity of
the armature insulation was increased
about ten per cent by operation in hydro
gen instead of air, while that of the field
insulation was practically doubled.
These increases in thermal conductivity
are practically independent of the hydro
gen pressure. The forced heat convec
tion from the internal machine surfaces
was estimated to be increased between
30 and 50 per cent by operation in hydro
gen at 0.44 pound per square inch pres
sure instead of in air. For increased hy
drogen pressures the forced heat convec
tion varies approximately with the 0.8
power of the absolute hydrogen pressure.
The heating data obtained in these
tests indicate that the 6,250-kva genera
tor as originally designed could have its
rating at 0.8 power factor increased by
very large percentages by operation in
hydrogen at various pressures instead of
in air, for the same temperatures of
the windings. However, these increases
in rating would be at the expense of elec
trical stability, as the short-circuit ratio
of a generator is decreased in the same
proportion as the output is increased.
W A T T S P E R SQ I N O F A R M A T U R E COIL
To obtain the same stability from the hy
drogen-cooled generator as from the air- Figure 5. Armature heating characteristics of
cooled one, the former must be propor a 6,250-kva generator with respect to tem
tioned somewhat differently. For the perature of gas leaving coolers, from tests with
6,250-kva generator redesigned for the generator operating in different cooling gases
SnellHydrogen-Cooled Generators ELECTRICAL ENGINEERING

0 0.1 0.2 03 0.4 0.5
WATTS PER Sq IN OF A R M A T U R E
.
COIL
0.7
Figure 6. Temperature drop from armature
copper to between-coil temperature detector,
for different watts per square inch of armature
coil, from tests on a 6,250-kva generator
operating in air and in hydrogen
gen pressures of 0.44, 15, and 25 pounds
per square inch, respectively.
Other Cooling Gases
Tests on the 6,250-kva generator using
carbon dioxide as the cooling agent indi
cated, as shown in figures 4 and 5, that
the heating of a machine in this gas is
practically the same as with air cooling.
Although the density of carbon dioxide
is about 50 per cent greater than that of
air and its thermal conductivity is about
JANUARY 1940, V O L . 59
35 per cent less, these disadvantages are
compensated for, with respect to the
heating of the windings, by the fact that
the forced heat-convection coefficient is
about 20 per cent greater in carbon di
oxide than in air, and the product of spe
cific heat and density is 39 per cent greater
for carbon dioxide than for air, so that 39
per cent more heat can be absorbed by a
given volume of the gas for the same tem
perature rise. Although there is no ad
vantage in the use of carbon dioxide as the
cooling agent for generators, as its higher
density greatly increases the windage
losses, there appears to be a distinct field
for its use in the case of induction motors
intended for operation in explosive at
mospheres. In the past year the General
Electric Company has placed in service
two induction motors of this type, using
carbon dioxide as the cooling agent, the
motors being provided with shaft seals
similar to those used with hydrogen-
cooled generators.
The use of helium as the cooling gas for
rotating machinery has been frequently
proposed, and has the advantage of ef
fecting a reduction in windage and ven
tilation losses comparable with that ob
tained with hydrogen, without the attend
ant disadvantage of inflammability in
mixture with air, possessed by hydrogen.
However, as shown by the curves in
figures 4 and 5, the cooling effect with
helium is distinctly inferior to that ob
tained with hydrogen, so that the in
crease in output of a generator by using
Figure 7. Longitudinal cross-section view of
a 55,555-kva 3,600-rpm hydrogen-cooled
generator
SnellHydrogen-Cooled Generators
helium in place of air as the cooling agent
would be considerably less than that ob
tained by the use of hydrogen in place of
air.
Generator Construction
The type of construction employed
with most of the hydrogen-cooled genera
tors built by the General Electric Com
pany is illustrated in figure 7 and in
cludes a cylindrical, gas-tight, outer cas
ing, reinforced with a number of annular
supporting-plates. Four gas coolers are
usually provided, located within the cas
ing with their long axes parallel to the
horizontal axis of the machine. The
cooler tubes are made accessible for clean
ing without removing the hydrogen from
the casing by removing cover plates at
either end of the frame. In another de
sign two coolers are used, placed length
wise of the generator and located in either
the top half or the bottom half of the cas
ing.
The outer end-shields of the generator
are of either cast-steel or fabricated con
struction and are usually employed to
support the generator bearings, thus
minimizing the distance between bearing
centers. The joint surfaces between the
end-shield halves and between the shields
and the casing are carefully machined,
and are provided with grooves within the
bolting line into which a plastic compound
is forced, thereby minimizing the possi
bility of hydrogen leakage from the
joints
The frame construction is designed to
withstand safely the explosion pressure
of the most explosive mixture of hydrogen
and air, with the generator operating at a
MMHHPMH
TRANSACTIONS 37

hydrogen pressure of 15 pounds per square
inch. This explosion pressure is esti
mated to be approximately 100 pounds
per square inch.
The rotors of 3,600-rpm hydrogen-
cooled generators consist of solid, one-
piece steel forgings, having slots milled
out of the body portion for carrying the
field winding. For many of these genera
tors the field winding is made from alumi
num, instead of from copper, which, due
to the lower specific weight of aluminum,
38 TRANSACTIONS
Figure 8. A 53,-
OOO-kw 3,600-rpm
steam turbine-gen
erator set; generator
hydrogen cooled
permits a larger rotor diameter to be em
ployed than with the copper winding and
thereby secures a greater rigidity for the
rotating shaft. The low rotational loss
in hydrogen permits such an increase in
rotor diameter, which would not be prac
tical with air cooling.
The ventilation of these generators, in
all sizes but the very largest, is provided
by two centrifugal fans, mounted one on
either end of the rotor. For the 176,470-
kva 1,800-rpm hydrogen-cooled generator
SnellHydrogen-Cooled Generators
of the Chicago District Electric Generat
ing Corporation, State Line, Ind., ven
tilation is provided by four motor-driven
fans, housed in domes in the top of the
casing.
A device is connected to the bottom of
the generator casing which operates an
alarm if water or oil should start to collect,
as could result from a defective cooler-
tube or from improperly-functioning
shaft seals.
Figure 8 shows a 58,889-kva 3,600-rpm
General Electric hydrogen-cooled genera
tor, installed in the Waterside station of
the Consolidated Edison Company of
New York. This machine has four gas
coolers, located two above and two below
the floor line. Distilled water is circu
lated through the coolers, and a heat ex
changer external to the machine, sup
plied with raw water, is used to cool the
distilled water.
The 50,000-kva 3,600-rpm hydrogen-
cooled generator of the Appalachian
Power Company at Logan, W. Va., em
ploys a novel cooling system in which the
stator core is cooled by direct contact of
the laminations with a number of hollow
metal pads, located between the lamina
tion packages, through which distilled
water is circulated.8 The construction
of this machine is illustrated in figure 9.
The rotor, and the armature end-wind
ings are cooled by the circulation of hy-
Figure 9. Longitudinal and end cross-section
views of a 50,000-kva 3,600-rpm hydrogen-
cooled generator having water-filled cooling
pads in the stator core
urtuztte coouite
ELECTRICAL ENGINEERING
drogen over them, the heated hydrogen
being cooled at the center of the machine
in passing to the back of the core through
ducts formed between special finned pads
and the laminations. A heat exchanger
external to the generator is used to cool
the distilled water circulated through the
pads. While no other generators of the
pad-cooled type have since been built,
this cooling arrangement is believed to
have several advantages which may some
time encourage its more general use. The
use of cooling pads in place of the custom
ary gas coolers permits a more compact
construction and reduces the volume of
gas that needs to be circulated.
Operating Hydrogen Pressure
The hydrogen-cooled generators built
by the General Electric Company are de
signed to operate normally at a hydrogen
pressure of approximately one-half pound
per square inch. Several machines have
been built, however, with provision for
operating at hydrogen pressures up to 15
pounds per square inch, if such operation
should become desirable for securing addi
tional kilovolt-ampere output or lower
generator temperatures under abnormal
water-temperature conditions. Operation
at hydrogen pressures greater than about
1.5 pounds per square inch requires the
provision of additional control features in
the seal-oil and gas control systems.
Shaft Sealing System
Characteristics
SEALS SUPPLIED WITH
VACUUM-TREATED OIL
The equations for the shaft sealing sys
tem employed with these generators are
derived in the appendix. These show
that, with vacuum-treated oil supplied
to the shaft seals, the degree of hydrogen
purity maintained in the generator casing
is a function chiefly of the degree of vac-
100
9
z Barometer 30 inches.
07
H
UJ
94i
15 20 25
TIME IN DAYS
JANUARY 1940, V O L . 59
uum treatment of the oil and the amount
of leakage from the casing. The purity
is shown to change exponentially with
time, following the initial starting-up of
the generator, finally reaching a constant
value. The amount of hydrogen con
tinuously required by the generator is
shown to equal approximately the casing
leakage plus the amount of hydrogen ab
sorbed by the sealing oil.
Figure 10 shows the calculated varia
tion in hydrogen purity with time follow
ing the starting-up of the generator, for
different degrees of vacuum treatment of
the sealing oil, for the 50,000-kva 3,600-
rpm Logan generator. A leakage from
the casing of 0.27 cubic foot per hour has
been assumed, based upon tests. The
curves show that several weeks of opera
tion are required for the hydrogen purity
in the casing to reach the final value corre
sponding to the vacuum and leakage
also that the final value is independent of
the initial hydrogen purity in the casing.
Figure 11 shows the calculated final
hydrogen purity in the generator casing
for different degrees of vacuum treatment
of the sealing oil and different assumed
values of the casing leakage, for the Logan
generator. A curve showing the estimated
yearly cost of the windage and ventilation
losses of this machine for different values
of hydrogen purity in the casing is also
given, which illustrates the economic ad
vantage of operating at a high purity.
This curve was calculated for an assumed
power cost of 0.3 cent per kilowatt-hour
and an operating time of 80 per cent.
It can be shown from the curves of figure
11 that for values of vacuum treatment
less than about 29 inches, a slight casing
leakage may increase the hydrogen purity
Figure 10. Calculated per cent hydrogen
purity in casing of a hydrogen-cooled genera
tor versus time for different degrees of vacuum-
treatment of oil supplied to shaft seals
Final hydrogen pur
ity by test 99 per
cent for 29.7 inches
vacuum in vacuum
tank. Hydrogen
consumption by test
1.48 cubic feet per
hour. Hydrogen
pressure in casing 10
inches water
SnellHydrogen-Cooled Generators
sufficiently to effect a saving in cost of
windage loss comparable with the cost of
the hydrogen wasted. In practice, how
ever, it is preferable to maintain a high
hydrogen purity by maintaining a high
degree of vacuum treatment, rather than
by leakage, and to have the casing leakage
as small as possible.
A high hydrogen purity is, of course,
essential from the standpoint of safety,
as well as of economy, of operation, since
30 26 26 24 22 20 16
DEGREE OF VACUU/ TREATMENT-IN.HG
600 600 1000 1200 1400 1600 1600
COST OF WINDAGE LOSS PER YEAR. - DOLLARS
Figure 11. Calculated per cent final hydrogen
purity in casing of a hydrogen-cooled genera
tor for different degrees of vacuum-treatment
of oil supplied to shaft seals and different
rates of hydrogen leakage q. Also estimated
yearly cost of windage losses versus per cent
hydrogen purity
Hydrogen pressure ten inches water
a mixture of less than 72 per cent hydro
gen in air is explosive. In practice it is
usual never to allow the hydrogen purity
in the generator casing to decrease below
90 per cent.
The degree of vacuum treatment of the
sealing oil is determined by the effective
ness of the treating system. For the
spray-recirculation system of vacuum
treatment used with these generators, the
vacuum treatment obtained is generally
between 0.1 and 0.3 inch less than the
vacuum held in the vacuum tank.
SEALS SUPPLIED WITH UNTREATED OIL
If the shaft seals are supplied with un
treated oil, as may occur under certain
emergency conditions of operation, the
air given up from the oil flowing to the
hydrogen side of the seals will cause the
hydrogen purity in the casing gradually
to decrease. This will require that the
casing be scavenged at intervals with
fresh hydrogen to keep the purity always
above 90 per cent. Figure 12 shows the
calculated and test values of hydrogen
purity under this condition of operation,
for the Logan 50,000-kva generator.
TRANSACTIONS 39
The curves show that, for the normal GENERATOR END SHIELD
rate of oil flow to the seals, with untreated (UPPER HALF)

oil, it is necessary to scavenge the genera ELECTRICAL

tor casing with fresh hydrogen about once CONNECTIONS

every 15 hours, to maintain the purity be

tween 90 and 95 per cent. The amount
of hydrogen of 99.8 per cent purity re GENERATOR BEARING
quired to increase the hydrogen purity in
PRESSURE REGULATING
the casing from 90 to 95 per cent is, ap VALVE SEAL OIL FEED

proximately,7
|A S^
CHECK
S
VALVE
PRESSURE-CONTROL
SWITCH
99 890 OIL
0.7X2.3X 1,000 log -500 cubic feet , , -SOLENOID
VALVE COOLER I
99.8-95 |>-
^ BEARING OIL
where the coefficient 0.7 is an "experience FEED
PRESSURE k
factor" and 1,000 the casing volume. MACHINE PRESSURE CONNECTION-
REGULATING
SEAL PRESSURE CONNECTION- VALVE
As this amount of hydrogen is required
every 15 hours the average hourly hydro
VENT
SEAL OIL DRAIN
VACUUM
TANK>
HYDROGEN DETRAINING TANK
N ^
gen requirement for this condition of
operation would be 500/15 = 33.3 cubic
feet per hour. This compares with 1.48
cubic feet per hour continuously required
by this machine to maintain a hydrogen
TO ATMOSPHERE
purity of 99 per cent when the vacuum
system is in operation. (The amount of \7~= END :
(LOWER HALF)
V ACTOR
hydrogen continuously required to main 1L>E ^:
tain a constant pressure in the casing,
when the seals are supplied with un
treated oil, is usually negative, that is, gas
must be vented from the casing. This MAIN TURBINE
OIL TANK ! SCREEN!
follows from the fact that for small values
of casing leakage more air is brought into
the casing by the sealing oil than hydro
gen is discharged by absorption and leak the oil the gas solubilities should be taken Figure 13. Arrangement of shaft-sealing sys
age.) as about one-tenth their values for the tem for hydrogen-cooled generator, with detail
rotating condition. With untreated oil of shaft-sealing arrangement shown for one
STANDSTILL OPERATION OF SEALS generator bearing. Arrangement for other
supplied to the seals, the gas solubilities
bearing is similar
The characteristics of the shaft-sealing used in the seal equations should be taken
system with the generator at rest are as about one-tenth their values for the
somewhat different than with the genera rotating condition. The practical results
tor operative, owing to the different of these differences in seal characteristics about equal to the casing leakage, and
values for the gas solubilities in the oil to for the stationary and rotating conditions the hydrogen purity decreases only about
be used in the seal equations. With vac are, that with vacuum-treated oil supplied one-tenth as rapidly as with the shaft ro
uum-treated oil supplied to the seals, two to the seals the hydrogen requirement is tating.
different values for both the hydrogen and only slightly less with the shaft at rest
the air solubilities must be used; for the than with the shaft rotating, but the hy Shaft Sealing System
absorption of gas by the vacuum-treated drogen purity in the casing is slightly Arrangement Details
oil, about the same solubilities may be higher. With untreated oil supplied to
assumed whether the shaft is rotating or the seals, however, the amount of hydro VACUUM-TREATING SYSTEM
stationary, while for the release of gas by gen required with the shaft at rest is The arrangement of the shaft-sealing
system employed with General Electric
hydrogen-cooled generators is illustrated
CUriVE fa)CALCULATED AND TEST VALUES OF P U f t l T Y FOR
Figure 12. Calcu diagrammatically in figure 13. The seal
Q!= 4 0 CU FT/MR (1.6 X NORMAL) lated and test values ing ring, which is attached to the genera
CURVE(b)CALCULATED VALUES OF P U R I T Y FOR Q i 3 0 of per cent hydrogen
C U F T / H R ( N O R M A L OIL FLOW) tor side of each bearing, contains an an
purity in casing ver
sus time, for a hydro
nular feed groove to which vacuum-
gen-cooled genera treated lubricating oil is supplied under
tor operating with pressure. The oil passes axially along
untreated oil sup the shaft in both directions from the feed
plied to shaft seals. groove to release grooves in the sealing
Casing scavenged ring and bearing, into which it is de
when purity de flected. The oil from the hydrogen-side
creases to 90 per release grooves passes into the hydrogen-
cent detraining tank, where the larger bubbles
Hydrogen pressure of gas absorbed by the oil flowing from
20
TIME
24
IN HOURS 18 inches water the seals are given up and pass back into

40 TRANSACTIONS SnellHydrogen-Cooled Generators ELECTRICAL ENGINEERING


Figure 14. Combination vacuum and hydro
gen-detraining tank with auxiliary equipment,
for shaft-sealing system of a hydrogen-cooled
generator
the generator. The oil from the air-side
release grooves unites with the bearing
oil and passes to the main oil tank, whence
an equal amount flows by gravity to the
air-detraining tank (or to an air-detrain
ing compartment in the main oil tank)
where the larger bubbles of air contained
in the oil are given up. From the hydro
gen-detraining and air-detraining tanks
the oil passes into the vacuum tank, where
practically all of the hydrogen and air re
maining in the oil is removed.
Oil is pumped to the seals from the
vacuum tank by a rotary pump of the
constant-displacement type, having a ca
pacity several times the normal seal oil
flow. Pressure regulating valve D holds
a constant differential between the seal
and hydrogen pressures by diverting into
the vacuum tank the part of the pump
delivery not sent to the seals. The oil
diverted through valve D enters the
vacuum tank through spray nozzles,
which serve to expose large areas of the
oil to vacuum treatment.
A vacuum of between 0.1 and 0.5 inch
of mercury, absolute, is held in the vac
uum tank by the operation of a rotary
vacuum pump. As the vacuum pump
discharges to atmosphere an explosive
mixture of hydrogen and air, the pump
is made explosion proof.
The by-pass tank shown in figure 13 is
used to divert the oil from the seal drain
to the air-detraining tank if it becomes
necessary to remove the hydrogen-de
training tank from service. Its cross-sec
tion area is sufficient to permit partial de
JANUARY 1940, VOL. 59
training of the hydrogen from the oil
under this emergency condition of opera
tion.
The principal elements of the vacuum-
treating system are usually constructed as
a unit, as shown in figure 14, and the
whole arrangement is conveniently lo
cated underneath the generator.
EMERGENCY SEAL OIL SUPPLY
In the event of any failure in the vac
uum-treating system which would cause
failure of the discharge pressure of the
seal oil pump, the shaft seals would be
supplied with untreated oil from the
bearing supply system through pressure-
regulating valve A, figure 13. This valve
is normally open but is prevented from
supplying oil to the seals by the normally-
closed solenoid-valve B. Failure of the
discharge pressure of the seal oil pump
opens the pressure switch C, allowing
valve B to open. This operation also
closes the normally-open solenoid-valve
E in the suction line of the vacuum pump,
preventing the rising of foam in the vac
uum tank, which would occur with the
spray nozzles unsupplied with oil.
With the shaft seals supplied with oil
from the bearing supply system, hydro
gen-saturated oil from the seals passes to
the air-detraining tank, and thence to the
main oil tank, through the overflow pipe
in the hydrogen-detraining tank. Opera
tion of the vapor extractor is relied upon
to maintain the atmosphere above the
oil in the main oil tank well below the
lower limit of hydrogen-air inflammability
(four per cent) under this condition of
operation.
ATMOSPHERIC RELIEF AND
SCAVENGING VALVES
The maximum operating hydrogen
pressure for generators not designed for
operation up to 15 pounds per square inch
pressure, is represented in figure 13 by
the elevation of the oil level in the main
oil tank above the low point of the over
flow pipe from the hydrogen-detraining
tank (dimension L). An atmospheric
relief valve of the mercury-column type is
provided at the hydrogen-detraining tank,
figure 15, which discharges hydrogen to
atmosphere if the pressure in the casing
approaches this limiting pressure. If
this valve should fail to operate, hydrogen
would be discharged into the vent pipe in
the air-detraining tank and would be
carried to atmosphere by the vapor ex
tractor.
A solenoid-operated scavenging valve is
also provided at the hydrogen-detraining
tank, figure 15, operated from the hydro
gen control cabinet, which allows hydro
SnellHydrogen-Cooled Generators
gen to be discharged to atmosphere if it
becomes necessary to increase the purity
of the hydrogen in the casing.
Hydrogen Control
CONTROL SYSTEM
The system of hydrogen control em
ployed with these generators is illustrated
in figure 15. The hydrogen supply is
contained in several commercial cylinders,
or "bottles/* connected with a manifold
through pressure-reducing regulators. A
solenoid-operated valve controlled by a
pressure switch admits hydrogen to the
casing from the connected cylinders to
maintain the casing pressure within pre
determined limits. These are normally
from 8 to 12 inches of water.
In filling the casing with hydrogen ini
tially, or in removing the hydrogen from
the casing, carbon dioxide is first ad
mitted, through a pipe at the bottom of
the casing, and the air or hydrogen dis
charged from the hydrogen feed pipe.
To determine when sufficient carbon di
oxide has been admitted to render the
mixture nonexplosive with hydrogen or
ifirWp
Figure 15. Arrangement of hydrogen piping
For hydrogen-cooled generator
air, an Orsat is used, or, in some cases, a
special scale on the hydrogen purity in
dicator is provided.
PURITY INDICATORS
An indication of the percentage hydro
gen purity in the generator casing is pro
vided by an instrument operating on the
thermal-conductivity principle. Hydro
gen from the casing is passed continu
ously through the analysis cell of the
purity-indicating instrument by the ac
tion of the generator fans or by continu-
TRANSACTIONS 41
 Table I. Heating Tests on 31,250-Kva 3,600-RPM Turbine Generator for Dayton Power and
Light Company Operating as a Synchronous Condenser at Zero Power Factor Overexcited

A l l Temperatures In Degrees Centigrade

Cooling Medium
Air Hydrogen

Kilovolt-amperes at zero power factor 19,200 30,000

Field input, kilowatts 62.5 98
Temperature of gas to coolers 63.4 46.1
Temperature of gas from coolers 40.3 33.1
Temperature rise of gas through coolers 23.1 13.0
Temperature of water to coolers 15.6 18.3
Temperature of water from coolers 26.5 26.6
Temperature rise of water in coolers 10.9 8.2
Water flow to coolers, relative 1.41 1.0
Temperature rise field winding above ingoing gas 67 67
Temperature rise field winding above ingoing water 91.7 81.8
Temperature rise maximum armature temperature coil above ingoing gas 38 32
Temperature rise maximum armature temperature coil above ingoing water 62.7 46.8

Figure 16. Instrument panel of hydrogen-
control cabinet for a hydrogen-cooled
generator
ally wasting a small amount of hydrogen.
In the analysis cell a platinum coil, ar
ranged in one of the arms of a Wheat-
stone Bridge circuit, is passed over by
the hydrogen at a slow rate. This coil
carries a heating current, supplied by a
12-volt battery, which causes the coil to
increase in temperature, and therefore in
resistance, by an amount depending upon
the thermal conductivity of the gas sur-
rounding it. Since the thermal conduc
tivity of hydrogen is about seven times
that of air, a small percentage of air in the
hydrogen will cause a relatively large
change in the thermal conductivity of the
hydrogen and will thus effect a propor
tionately large change in the coil resist
ance. The purity indicator is a galva
nometer-type of instrument, and indicates
the purity of the gas in the analysis cell
by comparing the resistance of the above
coil with that of a standard coil arranged
in another arm of the Wheatstone bridge
circuit. A contact in the indicator case
operates an alarm to indicate low hydro
gen purity.
A secondary indication of the hydrogen
purity in the generator casing is provided
by the reading of a gauge showing the
pressure developed by the generator fans.
As the fan pressure, with the generator
operating at constant speed, is propor
tional to the hydrogen density, the scale
of this gauge can be calibrated to indicate
percentage hydrogen purity.
CONTROL CABINET
The gas admission device, purity indi
cating equipment, and the other indicat
ing instruments required for operating
the generator in hydrogen are housed in a
control cabinet, usually located near the
generator on the operating floor. Figure
16 shows the instrument panel of the con
trol cabinet for a 55,555-kva 3,600-rpm
hydrogen-cooled generator. The large
round instrument shown at the center of
the upper row of gauges is the fan-pres
sure gauge. Directly below it is the hy
drogen purity indicator. Other instru-
Figure 17. Schematic diagram of electrical
circuits of the control cabinet for a hydrogen-
cooled generator. Items with (*) are located
away from cabinet

110 VOLT 1 CONTACT MAKING

RELAY N0.1

SWITCH

TJOy '
SOLENOID 6AS
ADMISSION
VALVE
110 VOLT A - C '
SOLENOID'
VALVE 125 VOLT D-C PILOT
EMERGENCY
SEAL O I L * PILOT LIGHT U 6 H T 125V0LT0-C

S O L E N O I D VALVE VACUUM PUMP SUCTION

42 TRANSACTIONS SnellHydrogen-Cooled Generators ELECTRICAL ENGINEERING

ments shown are the gauge for indicating
the hydrogen presssure in the generator
casing, the gauge for reading the reduced
hydrogen-bottle pressure, a gauge for
reading the vacuum in the vacuum tank,
and a dial thermometer for indicating the
hydrogen temperature at discharge from
the coolers. The different instruments
are provided with contacts which operate
the alarm system of the cabinet to indi
cate any abnormal condition. The cabi
net is mounted on rubber supports to pro
tect the instruments against vibration.
The three hand-valves shown at the
bottom of the cabinet in figure 16 corre
spond with valves 1, 2, and 3 in figure 15.
The alarm system for the control cabi
net is represented in figure 17 and includes
two contact-making relays of the vacuum-
tube type which are used to operate two
alarm horns. Operation of the alarm
contact of any instrument lights a signal
lamp and causes a current to flow through
a small step-up transformer; this im
presses a grid-bias on the vacuum tube
of one of the contact-making relays,
causing the relay to drop out and operate
the horns. This arrangement permits
a minimum passage of current through
the instrument contacts for the operation
of the horns.
Operating Experience with Eight
Hydrogen-Cooled Generators
Within the past 18 months eight Gen
eral Electric hydrogen-cooled generators
in capacities ranging from 31,250 kva at
3,600 rpm to 176,470 kva at 1,800 rpm
and with a combined output of 522,164
kva, have been placed in service. The
individual ratings of these machines, and
their service records, are given in figure
18. These units have operated almost
continuously since installation, with no
outages from faulty operation of the hy
drogen features, and have amply dem
onstrated the many advantages of hydro
gen cooling. The rates of hydrogen con
sumption have averaged from 35 to 60
cubic feet per day for the six 3,600-rpm
units, and are about 65 and 120 cubic
feet per day respectively for the two
1,800-rpm units. The costs of these
amounts of hydrogen range from $0.45 to
$1.50 per day. The hydrogen purities
Figure 18. Ratings
and lengths of time
in service of eight
General Electric
hydrogen - cooled
generators OCT NOV. DEC. JAN. FEa MAR.APR. MAY JUNE JULY AUG. SEPT. 0CT NOV. DEC. JAN. FEa MAR. APR.
1*37
JANUARY 1940, VOL. 59
in the casings of these machines have
been maintained at values between 97.5
and 99 per cent.
Although some little time is usually re
quired to familiarize the operators with
the various details of the auxiliary equip
ment no real operating difficulties have
been encountered, and in most cases the
operators have come to regard the opera
tion of hydtogen-cooled generators as
being only little more complicated than
the operation of the air-cooled type of
machine.
Heating Tests on a
31,250-Kva Generator
The first hydrogen-cooled generator to
be placed in commercial service, the 31,-
250-kva 3,600-rpm Dayton machine, was
given extensive tests in hydrogen and in
air previous to shipment. Some of the
test results are shown in table I and figure
19. It will be observed from table I that
this machine, operating at zero power
factor, carried in hydrogen approximately
NEW ORLEANS - 3600RPM:37,500KVA
ROCHESTER - 3600RJ?M.- 3I.250KVA
CINCINNATI 1800RPM.-81,250KVA.
CONS. EDISON 3600RPM.-58,889KVA.
PUB. SER. OF N. J.-3600R.PM.-55^55KVA.
STATE LINE GENr 1800R.PMrl76,470KVA.
LOGAN 3600RPMr50,000KVA.
DAYTON 3600R.PM.-3I.250KVA.
93 1939
SnellHydrogen-Cooled Generators
Figure 19. Temperature rise of cooling gas in
ventilation circuit of a 31,250-kva 3,600-rpm
generator built for operation in hydrogen, from
tests in air and in hydrogen, with generator
operating as an overexcited synchronous-
condenser at zero power factor
Paths of ventilatins gas shown by arrows
56 per cent greater load than in air for the
same temperature rise of the field wind
ing, with respect to the temperature of the
gas leaving the coolers, while the arma
ture winding operated at a temperature
rise even lower. Operating with air as
the cooling medium, this generator could
carry nearly full load rating with the
same heating of the windings as for a
standard air-cooled machine. In hydro
gen the full-load temperature rises are
considerably lower than the normal
values for air-cooled machines. The
curves of figure 19 show the marked re
duction with hydrogen cooling, in the
temperature rise of the gas in the air gap
which has an important influence on
the heating of the rotordue to the lower
rotational loss in hydrogen.
Future Trends in Generator Design
It will be apparent, from the large num
ber of hydrogen-cooled generators now in
successful operation, that this new type
of machine has now definitely emerged
from the experimental stage and has be
come an accepted piece of central-station
equipment. Much remains to be done,
however, now that some operating experi
ence with these machines has been gained,
toward simplification of the equipment
and improvement of the general design.
In developing the design described in the
foregoing, the General Electric Company
has endeavored to obtain a system that
TRANSACTIONS 43
would b e as reliable a n d as nearly "fool
proof" a s possible, even a t t h e expense
of some complexity. Present develop
m e n t a l work is being directed t o w a r d t h e
reduction in t h e n u m b e r a n d size of t h e
auxiliaries a n d in t h e simplification of t h e
control e q u i p m e n t . A new t y p e of shaft
seal employing smaller shaft clearances
t h a n t h e present seal h a s been developed,
which p e r m i t s t h e use of smaller t r e a t i n g
t a n k s a n d e q u i p m e n t t h a n a r e used a t
present, a n d requires less hydrogen. A
new p u r i t y indicator, with hydrogen a n d
c a r b o n dioxide scales, h a s been developed,
which is more rugged t h a n t h e indicator
used on t h e earlier machines. A new
s t a t o r frame design permits greater ac
cessibility t o t h e coolers for servicing,
t h a n t h e present design.
A t present n o hydrogen-cooled genera
t o r s smaller t h a n 28,125 k v a a t 3,600 r p m
or smaller t h a n 81,250 k v a a t 1,800 r p m
h a v e been built, or are u n d e r construc
tion, b y t h e General Electric C o m p a n y .
F o r sizes appreciably smaller t h a n these,
t h e gains with hydrogen cooling d o n o t a t
present a p p e a r sufficient t o justify t h e
additional expense a n d complication in
volved in its use. However, further im
p r o v e m e n t s in design, a n d increasing
familiarity w i t h hydrogen cooling on t h e
p a r t of t h e i n d u s t r y m a y eventually m a k e
desirable t h e extension of hydrogen cool
ing t o include even t h e smaller sizes of
machines. F o r t h e larger sizes, t h a t is,
a b o v e a b o u t 37,500 k v a a t 3,600 r p m a n d
75,000 k v a a t 1,800 r p m , it is probable
t h a t practically all future generators will
b e hydrogen cooled.
Appendix
Equations of t h e Shaft Sealing S y s t e m
In the following are derived the equations
for thehydrogen consumption and thehydro-
( Q 1 + Q 2 A OF AIR
SEALING RING
I / ' Q l Hz OF Ha LOST
r
MAIN TURBINE
HYDROGEN f Q1+Q2
DETRAINING OIL TANK
TANK
Q1 Hz OF H 2 O-
Q,A(1-Z) OF AIR
Figure 2 0 . Schematic diagram of shaft-sealing
system for hydrogen-cooled generators, using
untreated oil
44 TRANSACTIONS
gen purity in the casing of a hydrogen-
cooled generator provided with shaft seals
of the oil-film type. These equations are
based upon the theory of this t y p e of shaft
seal as given in the article of reference 5.
(a) SEALS SUPPLIED W I T H
VACUUM-TREATED O I L
Referring to the schematic diagram of the
sealing system in figure 2, let
Qi = n u m b e r of cubic feet of oil flowing
into the generator casing per hour
(from both shaft seals)
Qt=number of cubic feet of oil flowing
out on the air side per hour (from
both shaft seals)
JC=number of cubic feet of hydrogen
contained in one cubic foot of oil
coming from the vacuum tank at
the time t
y = number of cubic feet of air contained
in one cubic foot of oil coming from
the vacuum tank a t the time t
q = number of cubic feet of gas leaving
the casing per hour as leakage a t
the time /
2, = number of cubic feet of hydrogen of
purity 5 entering the casing per
hour a t the time t
t = time in hours
i n s o l u b i l i t y of hydrogen in the oil
leaving the seal (that is, the number
of cubic feet of hydrogen in one
cubic foot of oil)
i 4 = solubility of air in the oil leaving
the seal (that is, the number of cubic
feet of air in one cubic foot of oil)
2 p e r unit hydrogen purity in casing
a t the time t
1 2=proportion of air in casing a t time /
Z 0 = per unit initial hydrogen purity in
drum (that is, z=Z0 a t / = 0 )
V=total volume of gas in drum and
connected apparatus in cubic feet
5 = p r o p o r t i o n of hydrogen in gas supply
per unit
15 = proportion of air in gas supply, per
unit
P = a b s o l u t e pressure corresponding to
degree of vacuum treatment of oil,
atmospheres
For the determination of the hydrogen
consumption, referring to figure 2, note t h a t
the amount of gas in the oil flowing to the
seals equals the difference between the
amount of gas in the oil entering the vacuum
tank and the amount of gas discharged
from the vacuum pump. This can be shown
to be
QxPWz+Att-zK+QtAP
which must equal
Qi(x+y)+Q2(x+y)
Equating the hydrogen terms and the air
terms in these two expressions gives
PHz
* = <2i (i)
and
QtPA(l-z) , Q2AP
y= ~ . ~+- ~ (2)
Ql+Qt Ql+Q2
For the condition of constant gas pres-
SnellHydrogen-Cooled Generators
sure in the generator casing the following
equation must be true
Qi(x+y) +Qs -q+QiHz+QiAH -z) 0)
Substituting the values of x and y from
equations 1 and 2 in (3) and solving for qs
gives
qs = q+Qi [Hz+A(l -)]-
-j^iQtlHz+Ail-zil+QtA] (4)
yi-r/2
cubic feet per hour
for the amount of hydrogen t h a t must be
continuously supplied to the generator cas
ing to make up for t h a t lost through absorp
tion in the sealing oil and t h a t lost by leak
age. For high degrees of vacuum and purity
both P and (1 2) are small so t h a t the terms
of equation 4 containing these quantities
may be neglected and the equation for the
approximate hydrogen requirement of the
generator written as
qs = q+QiHz cubic feet per hour (5)
T o determine the degree of hydrogen
purity z t h a t would be maintained in t h e
generator casing with the system shown in
figure 2, let the purity a t time t be 2, and at
time t-\-dt be z+dz. The latter purity can
be shown to be
z+dz = (6)
Substituting the value of x from equation 1
an expression for dz/dt is obtained of t h e
form
= C +T D (7)
dt V
the solution of which is
et
"V (8)
where
C=2+ [Cid-maix
Qi+Qi
[H-S(H-A)} (9)
and
D=~[q+QiA(l-P)] (10)
The second term in equation 8 becomes
zero for / = infinity, so t h a t the first term
represents the final hydrogen purity in the
casing. I n figure 10 the hydrogen purity in
the casing of the Logan generator, as a func
tion of time following the starting-up of the
generator, has been calculated from equa
tion 8 for different values of P , for the value
of q obtained in tests on this machine. In
figure 11 the final values of purity from equa
tion 8 have been calculated for different
values of P and q. In these calculations S
has been taken as 0.998. The rates of oil
flow from the gas and air sides of the seal
have been taken from tests as Qi = 30 and
Q2 = 61.6 cubic feet per hour, respectively.
An average temperature of the seal oil of
42 degrees centigrade was used, for which
the hydrogen and air solubilities were taken
as 4 = 0 . 1 1 and H=0.041, respectively.
The casing volume V was taken as 1,000
cubic feet.
ELECTRICAL ENGINEERING
(b) SHAFT SEALS SUPPLIED W I T H
UNTREATED LUBRICATING O I L
For this condition of operation figure 20
applies. For the condition of constant gas
pressure in the generator casing we have,
using the symbols given under (a),
qs+QiA=QlHz+QiA(l-z) (")
whence
qs = q Qiz(A H) cubic feet per hour (12)
for the amount of hydrogen continuously re
quired by the generator with the seals sup
plied with untreated lubricating oil. For
small values of q, qs is negative, that is,
hydrogen must be discharged from the cas
ing to hold a constant pressure.
To determine the hydrogen purity in the
generator casing under this condition, re
ferring to figure 20 the purity at any time
t+dt can be represented by the equation
Vz-{QiHz+qz-qsS)dt
z-\-dz (13)
whence
dz_-ZE
+F (14)
dt~~
The solution of equation 14 is
F I -FV\ _*.'
(W)
where
E= q+Ql[H+S{A-H)] (16)
and
-4
(17)
In figure 12 the hydrogen purity in the
casing of the Logan generator, with un
treated oil supplied to the shaft seals, has
been calculated from equation 15 for two
different values of oil flow to the hydrogen
side of the seals, namely, Q\=30 and Qi=48
cubic feet per hour. For (?=48 cubic feet
per hour, using an air solubility A =0.11 in
equation 15 gave values of purity z which
checked the test values of purity exactly.
This value of A was used, therefore, in
calculating curve b in figure 12 and the
curves of figure 10.
Solubilities of Air and
Hydrogen in the Sealing Oil
In applying the equations of the fore
going section, the solubilities of air and
hydrogen in the oil supplied to the shaft
seals must be either known or assumed.
The curves in figure 21 give the solubilities
of air and hydrogen in transformer oil, which
has a viscosity considerably lower than that
of turbine oil and should therefore absorb
gas more readily. Tests made by the writer
indicated the solubility of air in turbine oil
to be 10.8 per cent at 40 degrees centi
grade, which agrees closely with the curve
for transformer oil infigure21.
In the calculation of figure 12 in the pre
ceding section the assumption of an air
solubility of 11 per cent, for oil at 42 degrees
centigrade, was found to give transient
values of hydrogen purity in the casing of
the Logan generator which checked the
JANUARY 1940, VOL. 59
*-H 2 SOLUBILITY IN TURBINE Oil.
FROM LOGAN TESTS
"0 20 40 60 80 IOO
TEMPERATURE OF OIL - DE C
Figure 2 1 . Solubilities of air and hydrogen
in transformer oil at atmospheric pressure and
at the temperature of the determination, for
different oil temperatures, from tests b y Rod
man and Maude 9
test values exactly. This value for air
solubility also agrees closely with the curve
for transformer oil given in figure 21.
However, calculation of transient values of
hydrogen purity for the Cincinnati genera
tor indicated that an assumption of an air
solubility of seven per cent, for oil at 36
degrees centigrade, was required in order to
check the test values of hydrogen purity.
However, the turbine oil for the Cincinnati
machine had a viscosity of 200 seconds
Saybolt Universal at 36 degrees centigrade,
as compared with a viscosity of 130 seconds
Saybolt Universal at 42 degrees centigrade
for the oil in the Logan turbine. Also the
shaft speed of the Logan generator was 45
per cent higher than that for the Cincinnati
generator, so that there was greater likeli
hood that all the air in the seal oil of the
Logan machine would be given up to the
hydrogen in the casing than would be the
case for the Cincinnati generator.
It may be deduced from the above that,
for the solubility of air in the turbine oil to
be used in the equations for the shaft seal,
the curve for air solubility in transformer oil
in figure 21 may be assumed correct if the
oil is of low viscosity, and if the shaft speed
is of the order of 12,000 feet per minute;
however, for oils of high viscosity and at
fairly low shaft speeds, the air solubility
should be taken as 60-70 per cent of the
values for transformer oil.
For determining the solubility of hydro
gen in the turbine oil, tests made on the
Logan generator at standstill with the shaft
seals supplied with treated and with un
treated oil were used. Substituting the
test values of hydrogen consumption for
these two conditions in equations 4 and 11
of the preceding section gave a value for
hydrogen solubility of 0.037 at 30 degrees
centigrade. This is 62 per cent of the
solubility of hydrogen in transformer oil at
this temperature. Assuming this same
percentage to apply for other temperatures,
the hydrogen solubility at the running
temperature of 42 degrees centigrade was
SnellHydrogen-Cooled Generators
estimated to be 0.041 and this value was
used in estimating the curves in figures 10
and 11.
Until more definite information is avail
able regarding the solubility of hydrogen in
turbine oil under the conditions which ob
tain at the shaft seals of a hydrogen-cooled
generator, it may be considered sufficiently
accurate to assume this solubility to be from
60 to 70 per cent of the solubility for hydro
gen in transformer oil, as given byfigure21.
References
1 HYDROGEN AS A COOLING MEDIUM FOR ELEC
TRICAL MACHINES, E. Knowlton, C. W. Rice, and
. H. Freiburghouse. AI BE TRANSACTIONS, vol
ume 44, 1925, pages 922-34.
2. OUTDOOR HYDROGENVENTILATED SYNCHRO
NOUS CONDENSERS, R. W. Wieseman. AIEE
TRANSACTIONS, volume 48, 1929, pages 1221-9.
3. T H E APPLICATION OP HYDROGEN COOLINO TO
TURBINE GENERATORS, M. D. ROSS. AIEB
TRANSACTIONS, volume 50, 1931, pages 381-6.
4. HYDROGEN COOLING OF ROTATING MACHINES,
C. M. Laffoon. AIEE TRANSACTIONS, volume 55,
1936, pages 703-09.
5. LIQUID FILM SEAL FOR HYDROGEN-COOLED
MACHINES, C. W. Rice. General Electric Review,
volume 30, 1927, pages 516-30.
6. C. W. Rice, United States Patent No. 1,559,182.
7. INTERCHANGING HYDROGEN AND AIR IN HYDRO
GEN-COOLED MACHINERY, C. J. Fechheimer.
Electric Journal, volume 28, 1931, pages 361-2.
8. M. A. Savage, United States Patent No. 2,099,-
575.
9. EFFECTS OF GASES IN TRANSFORMER OILS,
C. J. Rodman and A. H. Maude. American Elec
trochemical Society Transactions, volume 47, 1925,
pages 71-92.
10. LIMITS OF INFLAMMABILITY OF GASES AND
VAPORS, United States Bureau of Mines Bulletin
No. 279.
11. HYDROGEN-COOLED GENERATORS, E. H. Frei
burghouse and D . S. Snell. Power, volume 82,
number 8, 1938, pages 38-41.
12. SOME DESIGN FEATURES OF HYDROGEN-
COOLED TURBINE GENERATORS, D. S. Snell.
General Electric Review, volume 42, number 2, 1939,
pages 78-80.
Discussion
Sterling Beckwith: See discussion, page 34.
E. H . Freiburghouse (General Electric
Company, Schenectady, N . Y . ) : The re
liability of hydrogen cooling of large tur
bine generators has been thoroughly demon
strated. Referring to figure 18 of Mr.
Snell's paper, it is seen that a continuous
service record without interruption from
any cause associated with the hydrogen-
cooling features has been obtained on all
of the generators of the design described in
his paper. Nine of these machines are now
in service. As a result of our operating
experience with these machines, however,
a number of simplifications and improve
ments have been found possible and these
are being incorporated in the designs of our
newer hydrogen-cooled generators. Some
of these design changes and novel construc
tion features will be mentioned in this dis
cussion.
Figure 1 of this discussion shows the type
of construction used with our latest hydro
gen-cooled generators of the 3,600-rpm
type, and the arrangement of the gas-
TRANSACTIONS 45
 " 3^9
y*
Hs J
' ;
RSj 5^;7: tl
control and shaft-sealing systems. In this
design the four gas coolers are located above
the floor line, with the water connections
at the extreme ends of the frame instead of
at the sides, as in the earlier design. This
construction permits easier cleaning of the
cooler tubes than with the former arrange
ment.
The arrangement of the gas control sys
tem shown in the above figure is appreci
ably simpler than the arrangement formerly
used. Wherever possible, valves with dia
phragm-packed stems are being used to
reduce the possibilities of leakage. The
solenoid gas-admission and scavenging
valves, used with our present machines,
have been replaced in the arrangement
shown in figure 1 with diaphragm-type
valves. This was done to avoid the leakage
46 TRANSACTIONS
Figure 1
experienced with the solenoid valves, and
also to eliminate the mercury relief valve.
The arrangement of the carbon-dioxide
equipment shown in figure 1 is similar to
that used with the machines now in service,
and includes heating equipment which per
mits the carbon dioxide to be admitted to
the generator casing as a gas. Carbon di
oxide in the liquid form may be used thereby
reducing the time for purging and eliminat
ing the heating equipment, however, some
what more carbon dioxide would be re
quired.
Figure 2
SnellHydrogen-Cooled Generators
Our laboratory has developed a new type
of hydrogen purity indicator for use with
hydrogen-cooled machines, which is more
rugged and also more accurate than the
indicator now used. This operates on the
thermal-conductivity principle, as does the
present indicator, but uses alternating in
stead of direct current, thus eliminating the
need for the storage battery and rectifier.
The horizontal arrangement of shaft-
sealing equipment shown in figure 1 is being
used with a number of generators now under
construction. This provides greater accessi
bility of the parts; however, a vertical ar
rangement of the treating tanks has been
used wherever it was more suited to the
station space. The different elements of
either arrangement may be mounted on a
common base and assembled as a unit.
ELECTRICAL ENGINEERING
 One of the more unusual designs for
hydrogen-cooled generators is that employed
with the 137,500-kva vertical compound
turbine generator set built for the Ford
Motor Company, which is illustrated in
figure 2 of this discussion. This generating
unit, which is now being installed at Dear
born, Mich., consists of two separate hydro
gen-cooled generators, each of 68,750 kva
at 1,800 rpm, which are mounted one on top
of the other. Each generator is provided
with a double armature winding, and the
windings of the two machines are paralleled
through connections at one end as shown in
the above figure. The casing of one gene
rator is connected with the casing of the
other through the conduits used for the
electrical connections between the two
armatures, so that the two casings are con
sidered as one in filling or emptying the
casings with hydrogen and carbon dioxide.
However, each machine is provided with its
own hydrogen purity indicator and a sepa
rate fan pressure gauge. Two gas coolers
are provided with each generator, those of
the upper generator being located in the
bottom of the frame, while those of the
lower generator are in the upper part of the
frame. The hydrogen control panel, item
11, figure 2, is located in the side of the frame
structure of the lower generator. A single
set of treating tanks is provided for the
shaft-sealing oil for both generators, also a
single seal oil pump and vacuum pump.
This paper clearly indicates that much
has been accomplished and important prog
ress is being made in regard to the cooling
of turbine generators by means of hydrogen.
Operating experience is being rapidly ac
cumulated, also further improvements in
design and materials may be expected. It
is, therefore, certain that hydrogen cooling
will be an essential feature of large high
speed machines which are to be built, and
generators of even larger power output per
shaft will be possible in the near future.
Carl J. Fechheimer (consulting engineer,
Milwaukee, Wis.): The desirability of
using hydrogen as a cooling medium in
large high-speed electrical machinery has
been recognized since the presentation of the
splendid paper by Knowlton, Rice, and
Freiburghouse in 1925. Having done con
siderable work some years ago in the de
velopment of gland seals, devices for control
and proportioning of parts when hydrogen
cooling was beginning to be discussed, the
paper by Mr. Snell is of especial interest to
me.
It seems to me that the fear of an ex
plosion has caused our engineers to be overly
cautious. If an inert gas is used properly
as an intermediary when changing from air
to hydrogen or the reverse, it should not be
necessary to burden the generator with the
additional expense attended with making it
explosion proof. Particularly with the
large diameters associated with the large
slower speed 1,800-rpm machines, it is es
sential to make the enclosing brackets with
thick walls, probably with ribbing, and to
curve them, so that they will withstand an
explosion safely. I would inquire of the
author whether consideration has been given
to the design which will be sufficiently strong
for ordinary operation only, and if so, by
what percentage the cost of a large 1,800-
rpm generator would thereby be reduced?
JANUARY 1940, VOL. 59
Also, if the cost would thereby be reduced
appreciably, and if the control equipment
and shaft seals were simplified, as suggested
by Mr. Snell, would it not be economically
desirable to build hydrogen-cooled turbine
generators for smaller ratings than recom
mended in the paper?
It is of interest to study table I on tem
peratures. The losses with air for 64 per
cent of the load with hydrogen are 78 per
cent higher as given by the gas tempera
ture rise, and 87 per cent higher as given by
the water temperature rise. The fairly
close check between the two methods of
computation is interesting. Undoubtedly,
the windage loss was a very large percentage
of the total loss. Even with considerably
higher I2R losses, the field temperature rise
above the ingoing gas temperature was the
same, but was ten degrees less with hydro
gen, if referred to the ingoing water tempera
ture. As it is the ingoing water that is the
ambient, why not take the water tempera
ture, and not the gas temperature, as refer
ence?
The armature-coil temperature rises are
even more favorable for hydrogen, they
being only 75 per cent as high (referred to
the ambient water), even though the PR
loss was approximately 2.4 times as great.
This considerable gain was undoubtedly
due chiefly to the relatively high heat con
ductivity with hydrogen, which reduced the
thermal drop from the copper to the iron,
the transverse drop in the iron packages,
and the heat transfer drop from the cooling
surfaces to the cooling medium. Has Mr.
Snell given consideration in the design to the
feasibility of using thicker iron packages,
and of increasing the current density in the
armature copper, than would be employed
in an air-cooled machine?
It seems from the test results that the
usual ratio of rating for hydrogen to air
cooling of 1.2 to 1.3 is entirely too conserva
tive. If the data for this machine are
typical of other hydrogen-cooled turbine
generators, even 50 per cent increase in
rating is conservative.
In figure 9 the use of diffusers to improve
the efficiency and increase the pressure of
the centrifugal fan is interesting, and was
considered many years ago. It was found
that propeller fans were more efficient, were
not so much restricted at entrance as centrifu
gal fans, and could be applied to larger
sizes of machines. It is worthy of note that
with the very high peripheral velocities of
the centrifugal fans, more pressure can be
developed than is necessary; the result is
that the fans are not operated at the most
favorable point on their pressure-volume
curves. Is it not likely, as shown in table I,
that the high gas temperature rise when air is
used, is largely due to the high fan pressure
and rather low fan efficiency? If so, per
haps the ratio of ratings would not be so
high (as suggested in the preceding para
graph) if the temperature rise of the air
when passing through the fans were reduced?
As indicated in figure 9, a number of
hollow metal pads between lamination pack
ages, were used for the water coolers. The
details of construction are not entirely
clear and I would inquire whether the pads
interfered with the flow of hydrogen, and
whether the thermal drops from laminations
to water were appreciably altered, as com
pared with the usual finned surface cooler.
Perhaps one thing that has helped to
SnellHydrogen-Cooled Generators
bring the construction of hydrogen-cooled
machines to the forefront, is that now we can
calculate quite closely the fans, pressures
and volumes, as well as the distribution of
the cooling medium throughout the machine,
which could not have been done a compara
tively few years ago.
It seems that today, instead of hydrogen-
cooled machines being hazardous, they are
actually the safest machines, because there
is no internal gas to support combustion,
there is negligible damage from corona, and
they are as clean after years of service as
when first placed in operation.
P. L. Alger (General Electric Company,
Schenectady, N. Y.): When hydrogen
cooling for electrical machinery was first
proposed, the chief objection to its use was
the fear of an explosion. Tests made to
determine the pressure developed in an
explosion of the most explosive hydrogen-
air mixture, initially at atmospheric pres
sure, indicated that the explosion pressure,
for a machine assembled with core and
windings, would not exceed 50 pounds per
square inch. The theoretical explosion
pressure, for no heat loss to the surround
ings, is 180 pounds. Since, however it is not
difficult to construct the frame of the
machine to withstand a force of 200 pounds
per square inch, it is apparent that there is
H
0 20 40 SO lOO
P E R C E N T O F H E L I U M IN H Y D R O G E N
Figure 3
very little to fear from an explosion in a
hydrogen-cooled machine.
Nevertheless, it is interesting to consider
other gases which could be used for cooling
that are not subject to these explosion risks,
slight as they are, thus promoting economy
in methods of installation and handling.
An obvious possibility is the use of helium.
The light weight of helium gives very low
windage losses, but its materially higher
cost and its inferior cooling properties make
it undesirable for use as a pure gas. It
therefore occurred to me that the properties
of a mixture of helium and hydrogen should
be investigated, in the hope of obtaining a
gas free from explosion risk, that would pre-
TRANSACTIONS 47
serve practically all of the advantages of
pure hydrogen. The results of a theoretical
study of this subject by Doctor Saul Dush-
man and D. S. Snell are presented in the two
accompanying figures.
Figure 3 of this discussion shows the limits
of inflammability of a mixture of hydrogen
and helium, also the thermal properties of
i II 1 I I I I o
O 20 40 60 60 WO
PERCENT OF HELIUM IN HYDROGEN
Figure 4
the mixture compared with air. Figure 4
shows the calculated effect on the explosion
pressure at the upper limit of inflammability,
of mixing helium with hydrogen. The
curves were calculated for an assumed heat
loss of 25 per cent. The amount of helium
shown to be required, and its cost, in order
to produce a significant reduction in the
explosion pressure, are so large, however, as
to make this procedure unattractive.
C. F. Hill and L. J. Berberich (Westing-
house Electric and Manufacturing Com
pany, East Pittsburgh, P a . ) : The operat
ing experience described in these papers on
hydrogen-cooled generators indicates t h a t
hydrogen will in the near future be the
generally accepted cooling medium for
machines of the larger ratings. I t should be
mentioned again that electrical insulation
is subjected to fewer deteriorating influences
when operating in hydrogen as compared
with air.
We should like to re-emphasize one or two
of these. The effects of oxygen, ozone, and
oxides of nitrogen, the last two being pro
duced only under corona conditions, are
completely absent. These phenomena
have been discussed in previous papers and
need be only mentioned here. But even
more important, is the problem of moisture
absorption by the insulation which can be
almost completely eliminated in the hydro
gen-cooled machine.
Here the only source of moisture is from
the oil supplied to the hydrogen seals. T h e
moisture is, of course, introduced into the
lubricating oil system by the turbine.
When moisture-laden oil is passed into the
seals, which run in the neighborhood of
100 degrees centigrade, a considerable
amount of water is vaporized into the ma
chine. This water vapor acts as a contami
nant to the hydrogen and increases the ap
48 TRANSACTIONS
parent hydrogen consumption. Mr. Snell
in his paper stresses the contaminating in
fluence of air from the seal oil, b u t it is our
experience t h a t moisture can also be a seri
ous diluent of the hydrogen.
This moisture problem has been elimi
nated by the combination of an oil vacuum
treating unit and a gas drying unit which are
described by Messrs. Ross and Sterrett.
The simple vacuum-treating unit reduces the
water content of the oil to less than 0.003
per cent by weight. With added complica
tions, the vacuum system could be made to
reduce the moisture content still lower.
But even a t very low concentrations, some
moisture is always vaporized from the oil
in passing over the seals which will gradually
build up moisture concentration in the ma
chine and this will eventually increase the
hydrogen consumption. This, together with
the possibility of failure of the oil vacuum
treating system, justifies the addition of the
exceedingly simple activated-alumina gas
dryer described by Messrs. Ross and Ster
rett.
We have studied one of these machines
and the results may be of interest. Actual
determination of the moisture content of the
gas in a 50,000-kw machine which had been
in operation for some time has been made.
With the vacuum-treating unit and dryer in
operation the moisture content was approxi
mately 0.01 gram per cubic foot of gas.
This corresponds to a relative humidity of
0.4 per cent a t machine temperature of
55 degrees centigrade (machine was only
partially loaded at time of test), a dew point
in the neighborhood of 25 degrees centi
grade, and volume concentration of water
Figure 5. Shaft seal
AIR
vapor of approximately 0.06 per cent. The
effectiveness of this system of moisture and
air removal is attested by the fact that the
hydrogen consumption for this machine is
only 20 cubic feet per day. A similar test
was made on the same machine with the
moisture elimination system not in opera
tion and a water content of roughly 1.0
gram per cubic foot gas was found. This
corresponds to a relative humidity approxi
SnellHydrogen-Cooled Generators
mately 25 per cent a t 55 degrees centigrade,
a dew point of 30 degrees centigrade and
volume concentration of water vapor of
4.8 per cent. Under these conditions the
hydrogen consumption was very high and
water was condensed from the gas a t the
coolers which were operating approximately
10 degrees centigrade below the dew point.
The necessity of removing moisture from the
oil going to the seals is thus indicated.
Returning again to the electrical insula
tion, the combined oil-treating and gas-
drying system so effectively removes mois
ture and air from the oil that the insulation
may be considered almost hermetically
sealed in an inert medium. The problems
caused by the effects of ozone, oxides of
nitrogen, moisture, and accumulation of
dirt are all eliminated. The performance of
the insulation in hydrogen should therefore
be far better than in air.
S. H. Mortensen (Allis-Chalmers Manufac
turing Company, Milwaukee, Wis.): The
authors of the papers on hydrogen-cooled
turbine generators have so admirably
covered the design and operating problems
incidental to hydrogen-cooling of turbine
generators, that I find nothing to add except
a description of the Allis-Chalmers com
pany's hydrogen-shaft seal and a discussion
of its effect upon certain faces of the genera
tor operation.
The Allis-Chalmers company's shaft seal
is shown in figure 5 of this discussion and
consists of a small runner, item 1, provided
on the generator shaft at each end of the
machine close to the main bearing, and
OIL DRAIN TO
- DETRAINING
TANK
- CUP LEATHERS
against each face of the runner a conven
tional babbitt thrust bearing, item 2, is held
by springs, item 3, permanently set to give
a pressure equal to about a third of the
normal capacity of the bearing. Around the
inner periphery of each thrust bearing, but
separated from it by a groove, is a continu
ous babbitt face about an eighth of an inch
wide. During operation this continuous
face provides the required seal, since it is
ELECTRICAL ENGINEERING
 separated from the runner by only the thick
ness of the oil film in the thrust bearing.
Oil under pressure is circulated around the
flange between the two thrust-bearing faces
to provide cooling and lubrication and to
oppose the pressure of the hydrogen at the
small clearance between the continuous
babbitt face and the flange. The oil pres
sure counterbalances the spring pressure so
that resulting bearing load is very low.
Cup leathers, item 4, are provided at the
outer periphery of the oil chamber to retain
the oil and provide flexibility.
The Allis-Chalmers type of seal is suitable
for operation with hydrogen pressures from
zero up to ten pounds per square inch with
out oil or seal adjustments.
As this type of seal provides an air-tight
joint, air or gas can be removed from the
generator by evacuation. This makes it
practicable to simplify the scavenging proc
ess described by authors to the extent of
eliminating the use of carbon dioxide and
offering advantages such as;
1. No carbon-dioxide or other inert gas is required
for scavenging generator.
2. Since interchange of hydrogen and air takes
place at one fifth of an atmosphere or less, only
about one fifth as much excess hydrogen above one
volume is required.
3. Extra pipes to drain all "pockets" of carbon-
dioxide are not essential.
4. Less time is required to empty or fill generator
with hydrogen.
5. The filling operation is simplified.
Operating experience, over a period of
months, further shows that the sealing oil
volume which gets into contact with the
generator hydrogen is so small that oil
treatment becomes unnecessary except for
its passing through a hydrogen detraining
tank before it is returned to the oiling sys
tem.
I t is expected t h a t additional operating
experiences will result in further simplifica
tions in the application of hydrogen as a
ventilating medium.
Philip Sporn (American Gas and Electric
Service Corporation, New York, N . Y . ) :
Early in 1935 studies indicated the desir
ability of installing a t Logan, W. Va., a
40,000-kw 3,(500-rpm high-pressure super
position turbine, in order most economically
to develop and increase the capacity of the
existing power plant a t this location. At
that time the manufacturer stated that
cooling of the rotor for a generator of this
size and speed could be accomplished only
by utilizing hydrogen ventilation. Follow
ing our extensive and successful experience
with seven large-capacity hydrogen-cooled
synchronous condensers then in service on
the American Gas and Electric Company
system, we unhesitatingly accepted hydro
gen cooling for this new generator unit.
In the case of synchronous condensers, the
bearings are entirely enclosed in the outer
shell, and therefore the problem of sealing
the shaft against gas leakage is not en
countered. However, extensive tests and
factory operating experience with the oil
seal at Schenectady over a period of years,
provided a basis for anticipating satisfac
tory performance from this shaft-sealing
arrangement in power-station service.
The Logan generator went into operation
as a hydrogen-cooled machine on December
JANUARY 1940, VOL. 59
2, 1937, and from t h a t time until February
28, 1939, the unit has run a combined total
of 232 days. The running time lost during
this period has not been occasioned by any
trouble or maintenance in connection with
the generator proper or its gas sealing sys
tem. N o difficulty of any great importance
has been experienced with the shaft-sealing
system and its accessory equipment. No
failure to maintain the seals has occurred at
any time. As of February 28, 1939, this
seal system had operated 308 days. We are
now maintaining a purity varying from 99
to 100 per cent and the gas consumption is
averaging approximately 40 cubic feet per
day. The delivered cost of this gas, includ
ing handling of the cylinders, amounts to
approximately 2 cents per cubic foot, so
that the yearly cost on this basis is less than
$300.
Mr. Snell describes in his paper the pad
cooling arrangement for direct water cool
ing of the stator. Plant condensate was
used as make-up for this pad liquid system.
In order to eliminate the possibility of air
pockets in the pads, it is necessary to ob
tain a high vacuum before filling with con
densate. In this respect some difficulty
was experienced initially but by circulating
the pad liquid for some time and by wasting
a portion from the overhead expansion tank,
it was possible to remove practically all of
the air still remaining in the pads. The
operation of the system has been entirely
satisfactory.
Mention is made of operation at gas pres
sures as high, as 15 pounds per square inch.
The Logan unit has been run for short pe
riods at these higher pressures with satis
factory performance of the shaft-sealing
system and without obtaining excessive
consumption of gas. We have had consider
able experience with the operation of hydro
gen-cooled synchronous condensers at pres
sures up to and including 15 pounds per
square inch. We hope to present detailed
operating data covering such performance
in a coming paper before the Institute.
D. S. Snell: I wish to thank Mr. Fech-
heimer for his interesting and valuable dis
cussion. Mr. Fechheimer has contributed
much to the development of hydrogen cool
ing and his comments on the subject are
welcomed. As to the saving effected with
hydrogen-cooled generators of the 1,800-
rpm type through making the machine
structure strong enough only for normal
operation, but not strong enough to with
stand an explosion, I believe that for the
type of construction employed with our
hydrogen-cooled generators no saving would
be obtained. For with a construction in
which the generator bearings are supported
in the end shields, a construction that is
rigid enough for ordinary operation is also
strong enough to withstand an explosion.
At the present time the cost differential be
tween the hydrogen-cooled and the air-
cooled machine is principally due to the cost
of the auxiliary equipment, and as this cost
does not vary greatly with different sizes of
machines, it is difficult at the present time
to justify the use of hydrogen cooling for
generators in sizes much smaller than the
minimum sizes mentioned in the paper.
Regarding Mr. Fechheimer's proposal
that ingoing water temperature rather than
ingoing gas temperature be used as the ref-
SnellHydro gen-Cooled Generators
erence for basing temperature rises for
hydrogen-cooled machines, we have given
considerable thought to this matter but
still feel that the use of the ingoing gas tem
perature offers the best basis for making
guarantees on this type of machine. The
use of the ingoing gas temperature as a basis
for temperature rise allows an accurate
comparison between different machines as
the true rise of the machine alone is ob
tained, whereas if the ingoing water tempera
ture were used as a basis, a designer could
skimp the design of the generator and use
an extra large cooler, obtaining the same
over-all temperature rise with this machine
as with a machine of more liberal design.
Obviously the machine with the lower tem
perature rise within itself is the more liberal
design.
Also, since it is easy to test the coolers and
the generator separately, there is little or
no reason to combine the two in one over
all guarantee, and especially is this true due
to the fact t h a t the generator is the major
portion of the investment and the more
critical.
The Institute a t the present time has
adopted the ingoing gas or air as a basis for
the ambient temperature on which rises
should be predicated, and we can see no
reason for the Institute's making two differ
ent standards for large rotating equipment
because one happens to be enclosed and the
other an open machine, as basically the
temperature rises, which are satisfactory
for one machine are applicable to the other
type.
Regarding the thickness of the stator iron
packages and the armature current density
in the hydrogen-cooled machine, both these
quantities have been increased over then-
values for the air-cooled machine.
Mr. Fechheimer's suggestion that the in
crease in rating of a generator by the use of
hydrogen should be considerably greater
than the 20 per cent mentioned in the paper,
based on the data given in table I, neglects
the fact that the hydrogen-cooled generator
must be proportioned differently than the
air-cooled machine, to obtain the same elec
trical stability. If this change in design is
considered it will be found that the increased
output per pound of material with hydrogen
will be of the order of 20 to 25 per cent.
Regarding the use of diffusers in conjunc
tion with the centrifugal fans employed with
these machines, while these are generally
provided with air-cooled generators they
have been omitted from many of the hydro
gen-cooled machines for the reason that fan
efficiency is relatively unimportant in a
hydrogen-cooled machine, while the reduc
tion in machine length obtained by omit
ting the diffusers is of value. This omission
results in a rather high windage loss and
temperature rise of the air when the ma
chine is operated in air, as Mr. Fechheimer
has observed in connection with table I;
however, since these machines are intended
principally for operation in hydrogen, this
feature is not considered important. As to
the relative merits of centrifugal and pro
peller fans, we recognize that while the
propeller fan has a slight advantage in
efficiency it also requires a greater amount of
axial space on the shaft for its support, thus
increasing the distance between bearing
centers.
Referring to Mr. Fechheimer's inquiry as
to the construction of the machine illus-
TRANSACTIONS 49
trated in figure 9, figure 6 of this discussion
illustrates the construction of the metal
pads which are employed for removing the
heat from the stator core and the cooling
gas. The pads of the plain type are used in
the end sections of the stator core and re
move the generated heat by direct conduc
tion to the cooled pad. The finned pads
are used in the center part of the core and
absorb heat both by conduction through
contact with the laminations, and by con
vection, through passage of the heated gas
from the air gap to the back of the core be
tween the fins. The finned pads take the
place of the surface gas coolers used with the
ordinary type of generator. The fan pres
sure required to ventilate this type of ma
chine is practically the same as is required
with the ordinary type, for the same volume
of gas circulated.
It is true, as Mr. Fechheimer suggests,
that the development of the hydrogen-
cooled machine has been accelerated by the
progress previously made in methods of
calculating fan performance and ventilation
characteristics, and also, I would add, in
methods of predetermining the thermal be
havior of electrical machines. For much of
this progress we are indebted to the re
searches of Mr. Fechheimer.
The shaft seal described by Mr. Morten-
sen possesses many desirable features.
Several years ago we tested a seal of a some
what similar type, in which a stationary
carbon ring pressed axially against a runner
carried on the shaft. Trouble was experi
enced, however, through air being pumped
into the generator casing between the ring
and runner by centrifugal force. In Mr.
Mortensen's seal this pumping action is pre
vented by supplying lubricant to the rub
bing surface under pressure.
While it is desirable to eliminate from the
sealing system as many auxiliaries as possi
ble, the elimination of the vacuum tank, as
suggested by Mr. Mortensen has certain
disadvantages. Without the use of vac
uum-treated oil in the shaft seals, the hydro
gen purity in the generator casing will de
crease, even if at a slow rate, and the opera
tor must manually scavenge the casing at
intervals in order to maintain the desired
hydrogen purity. If the oil is vacuum
treated, however, no scavenging is required.
Also, the vacuum treatment of the sealing
oil, aside from its value in the operation of
the hydrogen system, is of definite value
with respect to the operation of the bear
50 TRANSACTIONS
ings of the turbine generator set, since vac
uum treatment of oil removes impurities
and thereby prolongs its life. Considering
that the vacuum tank requires little space
and practically no servicing, it does not
seem that the elimination of this device
offers as many advantages as its retention.
The evacuation of the generator casing
before filling with hydrogen, or when empty-
Figure 6. Monel-
meial cooling pad
for hydrogen-cooled
generator, copper-
brazed
Arrows indicate
brazed joints
ing, as proposed by Mr. Mortensen, has
some advantages over the use of an inert
gas. However, to evacuate the casing re
quires that the generator be at rest, whereas
the inert gas may be introduced with the
generator rotating, which is frequently done
to save time.
Messrs. Hill and Berberick mention the
problem of contamination of the hydrogen
by moisture given up from the sealing oil as
being of equal importance to the problem of
contamination by air from the oil. While
this is probably true where the temperature
of the oil leaving the seals is in the neighbor
hood of 100 degrees centigrade, it is not the
case where the temperature of this oil is
approximately 60 degrees centigrade, as is
the case with the type of seal described in
my paper. Since the radial shaft clearance
of this seal is the same as that of the bearing
the energy loss in the oil film is low compared
with the oil flow so that only moderate oil
temperatures are obtained. This precludes
the possibility of moisture being vaporized
out of the oil into the generator.
Mr. Beckwith suggests the elimination of
the hydrogen relief valve. In the system
described in my paper a relief valve of the
mercury type is used to prevent the dis
charge of hydrogen into the air detraining
tank in the event of excessive pressure in
the casing. In our later designs a dia
phragm type of relief valve is used, which
serves also for scavenging. In our case this
valve can hardly be dispensed with.
The hydrogen requirement for these ma
chines has been found to vary approximately
with the square root of the hydrogen pres
sure, as stated by Mr. Beckwith, although
theoretically the part of the hydrogen re
quirement due to absorption in the sealing
oil should vary about directly with the
absolute pressure. The discrepancy is prob
ably due to a slight opening-up of the cas
ing joints at the higher pressures. Tests on
the Logan 50,000-kva generator showed a
hydrogen requirement of 40 cubic feet per
day at one-half pound pressure and 300
cubic feet per day at 15 pounds.
SnellHydrogen-Cooled Generators
Mr. Sporn describes operating experiences
with the Logan generator, which was one of
the first hydrogen-cooled generators to be
placed in service. It is gratifying to us that
this machine, which embodied two radical
departures from conventional design-hydro
gen cooling and pad coolingshould have
operated practically without trouble since
installation. Mr. Sporn's company has
been a pioneer in the use of hydrogen-
cooled machinery, having installed in 1928
one of the first hydrogen-cooled synchronous
condensers and in 1937 one of the first hy
drogen-cooled generators. His faith in this
new development has contributed largely
to its success and it is gratifying to us to
see that his confidence in hydrogen cooling
has been justified.
Mr. Alger mentions a theoretical study
made to determine the desirability of using a
mixture of helium with hydrogen to reduce
the explosion pressure. At the time this
investigation was made the cost of commer
cial helium was approximately six times the
cost of hydrogen. Since then the cost of
helium has been greatly reduced and it can
now be obtained for approximately twice
the cost of hydrogen. Even with this reduc
tion in cost, however, the use of helium as a
cooling agent is not considered desirable
since its cooling properties, as explained in
the paper, are considerably inferior to those
of hydrogen. Since the type of generator
construction and the shaft-sealing and gas-
control equipment would be substantially
the same whichever cooling medium were
used, and the helium-cooled generator would
be appreciably larger than the hydrogen-
cooled machine for a given output, it will be
evident that the use of helium is not eco
nomically advantageous.
Mr. Freiburghouse has mentioned some of
the recent improvements that have been
made in the design of the hydrogen-cooled
generator, also some of the details of con
struction of the 137,500-kva hydrogen-
cooled generating set for the Ford Motor
Company. This generator set has recently
been placed in service with hydrogen. The
shaft seals used with these generators are of
a different type from those used with our
earlier generators and consist of metal
rings bearing radially on the shaft, the rings
being supported in a housing attached to the
end shield. This type of seal requires less
oil than the type described in the paper, so
that smaller treating tanks and equipment
may be used.
Besides the Ford generators, two other
hydrogen-cooled generators have been
placed in service by our company since my
paper was written. We have now 12 of these
machines in operation, their combined rat
ing being 770,000 kva. The largest 3,600-
rpm machine is rated 66,667 kva while the
largest 1,800-rpm generator is rated 176,470
kva. A total of 19 hydrogen-cooled genera
tors are under construction at the factory,
with a combined output of 1,035,000 kva.
One of these units is rated 50,000 kva at
3,000 rpm and will be installed at Buenos
Aires. This is the first hydrogen-cooled
generator to be sold for foreign installation.
The growing popularity of this new type of
generator indicates that the former appre
hensions regarding hydrogen cooling have
been largely dissipated by the successful
operation of the units now in service and
that the many advantages of hydrogen cool
ing are becoming widely recognized.
ELECTRICAL ENGINEERING